Phencyclidine (PCP), an NMDA antagonist, has been shown to mimic some aspects of schizophrenia including positive, negative and cognitive symptoms. Previous studies in this laboratory have shown a selective reversal-learning deficit following acute PCP administration, a deficit that is attenuated by atypical, but not classical, antipsychotic treatment. However, acute PCP has limitations for modelling the chronic psychotic illness and persistent cognitive deficits observed in many schizophrenic patients. Therefore, the aim of this study was to examine the cognitive deficit induced by PCP over a longer term using a previously established operant reversal-learning procedure. Moreover, the efficacy of the atypical antipsychotics clozapine, ziprasidone and olanzapine to reverse the sub-chronic PCP deficit was compared with that of the classical antipsychotics, haloperidol and chlorpromazine. Female hooded-Lister rats were trained to respond for food using an operant reversal-learning paradigm. When animals achieved criterion of 90% correct responding they were treated with PCP (2 mg/kg) or vehicle twice daily for 7 days, and 7 days later tested for their cognitive ability. PCP induced a significant impairment in the reversal phase relative to the initial phase of the task. Acute ziprasidone (2.5 mg/kg), olanzapine (1.5 mg/kg) and clozapine (5 mg/kg) produced a significant attenuation of the impairment induced by sub-chronic PCP in the reversal phase. In marked contrast to these effects, acute administration of the classical agents haloperidol (0.05 mg/kg) and chlorpromazine (2 mg/kg) failed to significantly reverse the PCP-induced cognitive impairment. These data clearly demonstrate that sub-chronic PCP produces enduring and persistent cognitive deficits, effects that are significantly attenuated by atypical but not classical antipsychotics.